Supervised Machine Learning Applications on Meteors and Meteorites

As part of the industrial revolution 4.0 and advents in space exploration, Artificial Intelligence (AI) is being integrated with numerous space-related technologies. This paper presents supervised machine learning algorithms developed to detect and recognize meteors and meteorites. The United Arab Emirates (UAE) Meteor Monitoring Network (MMN) system observes daily meteors from sunset to sunrise. The system records meteors in addition to objects giving off light or movement. This, in turn, results in false data and consumes time during manual filtration. The dataset is structured on the available data, exceeding thousands of observations. A supervised model is trained on this dataset and is finally tested on images captured from the UAEMMN stations. The model is structured using an object detection algorithm on top of a Convolutional Neural Network (CNN). Similarly, another CNN supervised model identifies meteorites and distinguishes them from rocks. The dataset comprises images of meteorites from the meteorite collection at the Sharjah Academy for Astronomy, Space Sciences, and Technology (SAASST) and pictures of rocks. The study revealed that both models proved reliable algorithms for meteor and meteorite detection, reaching an accuracy above 85%.